Our validations cover a full range of credit scoring and decisioning models, including:
Application Scorecards (Consumer & Commercial)
Behavioral and Account Management Scorecards
Collections and Recovery Scorecards
Credit Line Management Models
Machine Learning and Hybrid Models
We perform a comprehensive, end-to-end assessment of your model, including:
Conceptual Soundness and Model Design
We evaluate the model’s purpose, structure, and development methodology, ensuring it aligns with the institution’s lending strategy, credit policy, and risk segmentation needs. For statistical models, we assess variable selection, binning, and transformations for logic, stability, and predictive power.
Data Quality and Variable Integrity
We review the source, sufficiency, and consistency of input data used in model development and implementation. This includes assessing the handling of missing values, outliers, and potential biases, as well as the mapping of bureau data and internal attributes.
Model Performance & Discriminatory Power
We test the model’s performance using standard validation metrics such as:
KS Statistic
Gini Coefficient
Population Stability Index (PSI)
Characteristic Analysis (IV, WoE)
Override and approval rate tracking
Our analysis includes both development sample backtesting and out-of-sample / out-of-time validation where available.
Decisioning Logic and Cutoff Strategies
We validate the logic and thresholds used in automated decisioning rules, including approvals, declines, referrals, and exceptions. We assess whether decision strategies are calibrated to portfolio goals, regulatory constraints, and risk appetite.
Fair Lending and Bias Testing
If requested, our validations include reviews for disparate impact and unintended bias across protected classes (e.g., race, gender, age), consistent with fair lending regulations. We assess both model inputs and outcomes for compliance and ethical soundness.
Model Governance and Documentation
We assess model documentation, development records, version control, and the model’s integration into your model risk management (MRM) framework. We also review policies and procedures for model monitoring, periodic revalidation, and exception handling.
Platform & Vendor Coverage
ValuRisk Partners supports validations across a wide range of environments, including:
Vendor platforms (e.g., FICO, Experian Decision Analytics,)
Custom-built SAS, R, Python, or SQL scorecards
Excel-based rule engines and hybrid approaches
